Overclock.net banner

1 - 18 of 18 Posts

·
Registered
Joined
·
174 Posts
Discussion Starter · #1 ·
I'm a bit stuck with a watercooling build but seeing as it's in regards to fans I thought I'd ask here first.

I'll be using a fair amount of fans (16 on rads, 2 intake front, 2 exhaust rear). And all of them PWM.

- At this point my first dilemma is what fans to go for. My main choices are the new EK Vandar fans (1850 RPM versions) or the BeQuiet! Silent Wings 2. Given the fair amount of aditional radiator space I've got I'd wager I can get away with the BeQuiet! fans lower performance trade-off in return for noise. The EK's are cheaper by 5 euros per fan though and can (Presumably) be painted. I like either for looks. Alternatives are Noctua NF F12 fans and NB eLoops. Not a huge fans of the eLoop looks and the Noctua fans won't be very quiet. While I don't mind a little more noise, I'm looking for a good noise to performance ratio.

- My initial plan was to daisy chain Swiftech 8-way PWM splitters and supply PWM signal to all fans this way. Additional power supplied through SATA. Doyle mentioned the PWM signal may or may not be sufficient. My current Mobo is an Asus Z87 Hero, to be replaced by either a Gigabyte Z97 UD5 Black edition or Asus Z97 WS. I've looked at specs on all of these boards (And more) and in general regarding PWM signal strength but came back empty-handed. Are there resources related to PWM signal or is hooking up 20 fans to a single PWM signal silly to begin with?

- If my plans should include a fan controller capable of performing what I want, which ones are recommended? I know the Aquaero 6 is capable of PWM control

Appreciate any advise!
 

·
Overclock Failed...
Joined
·
13,565 Posts
All your fans will be running at the same speed and that speed will be based on CPU temp. Is that OK for you?
You could "split the splitters" and have some connected to the CPU header (CPU temp dependent speed) and some connected to the motherboard header(motherboard temp dependent).
 

·
Registered
Joined
·
174 Posts
Discussion Starter · #3 ·
That would work fine for me. On the other hand, is it possible to have them run off of a sensor's temperature?

Edit; Water sensor, that is.
 

·
Overclock Failed...
Joined
·
13,565 Posts
Quote:
Originally Posted by Domiro View Post

That would work fine for me. On the other hand, is it possible to have them run off of a sensor's temperature?

Edit; Water sensor, that is.
On the surface one would think that would be a good idea. but it isn't.
When CPU temps fluctuate wildly (as they do) this causes fans to speed up and slow down quite unnecessarily as typically, it takes somewhere between 5 and 30 seconds of high stress on the CPU for water temps to rise, say 1°c, and perhaps 2 - 5 minutes until speeding up fans would actually be necessary. So, yeah, keying fan speed off water temp seems like a great idea.

But it isn't water temp that you want to maintain, it's Delta T, the temperature difference between air entering the rad(s) and the water in the loop. Controlling that temperature difference controls the efficiency of the whole WCing system. Low Delta T for high CPU loads and a high Delta T with quite, low fan speeds during low load conditions.

So, what is really needed are sensors on both the water and air and software the compares those readings and adjusts the fan(s) voltage accordingly to maintain a Delta T. As far as I know such stuff isn't available in the WCing industry. To further complicate matters, Delta T also varies based on ambient temps. A WCing system that runs at a Delta T of say. 5°c will run at a higher Delta T (8°c ??) with higher ambient temps, all else being equal. So, that hypothetical fan control software also has to compensate for ambient temps too.

All this is doable, and in fact is done in the HVAC industry (Heating Ventilation and Air Conditioning), but there is not, as far as I know, any practicable way to incorporate those sensors and control systems in to a WCing setup at anything near what one would consider even outrageous costs. Think in the range of $1.000 minimum. ...and then throw in pump speed control too!!!!

So, what are we to do?

Here's the way I do it after close to 15 years of water cooling including world rank benchmarking multi pump, multi rad, multi GPU systems:
I cool only the CPU as I'm not a gamer and have been all through GPU benchmarking already.
I use an Iwaki (MD20RZ(t) pump and three Delta 1212EHE fans on a 3 x 120 BIX rad and a Swiftech Storm block (the "Cather style" impingement one).
I run the fans at 5 volts which gives me great Delta Ts in winter and passable Delta Ts in the summer (sans AC).
When I stress test or benchmark I run the fans for 12 volts (yeah, they sound like three vacuum cleaners!) and get fantastic Delta Ts no matter what the ambient is.

Your 16 fans at low voltage is going to take care of pretty much any everyday load and just crank then up (with a fan controller, rheostat or 5v / 12v switch) if you need to.

Sorry be being so long winded ....but you did ask!
 
  • Rep+
Reactions: Domiro

·
Registered
Joined
·
174 Posts
Discussion Starter · #5 ·
Hahaha I did ask indeed!

I fell asleep thinking "What about Delta T?" and I see now why it would be difficult enough to maintain. Your first suggestion then makes sense: two sources of heat in the loop being the CPU and 2 GPUs. Using half the batch of fans on one for CPU with a custom fan curve and the second batch with a different curve and sensor. I could go as far as use an Aquaero6 with its seperate channels (And depending on fan choice. I'm going to have a look around for fan controllers, see what suits my needs best. It'll influence what fans I pick for the setup.

Thanks for the thorough explanation
 

·
Premium Member
Joined
·
27,533 Posts
The critical temps are WATER TEMP LEAVING CPU and GPU blocks. This is the hottest water. Knowing / using this temperature source is best water temp control as it is both dependent on water temp going into waterblock and heat coming from CPU / GPU. If radiator fans and pump/s speed up as water them out of block rises, the fans will be speeding up as water reaches cooler.

Let's not over-complicate the cooling system. Delta temp means nothing to the cooling system. Cooling increases as needed by component. Its' relevance is for us to know how efficient system is, but is not part on how system implements cooling.

Water cooling is how most vehicle and heavy equipment engines are cooled. I've yet to see one that uses air temp. System increases cooling as need to keep engines within operating temperature range. The hotter the engine the more cooling (fan speed / airflow) is used. Increased air temp only means more cooling ability is needed to keep engine cool. None that I know of use air temperature, only engine temperature. Same principle applies to computer components.
 
  • Rep+
Reactions: Domiro

·
Overclock Failed...
Joined
·
13,565 Posts
Quote:
Originally Posted by doyll View Post

The critical temps are WATER TEMP LEAVING CPU and GPU blocks.
If you use water temp (only) to control fan speeds the fans will speed up and slow down based on ambient temps, as that's what the water does. On cool days your fans will run slower and on warm days you fans will run faster at the same CPU/GPU loads. Not really what we want. Controlling Delta T controls how much heat is actually being removed.

If you made a chart that relates Delta Ts to CPU/GPU temps at a constant load and varying ambient temps (using IBT, as it more accurately represents real world full loads as compared to Prime95 or OCCT) and use that to manually set fan speeds, then fans would always run at the minimum speed required to supply the amount of cooling needed.
But that's way to complicated and basically, the only thing it really accomplishes is:
1.) To save a small amount of electricity (compared to just changing fans from 5v to 12v as needed)
2.) To run the fans quieter (sometimes).
 

·
Premium Member
Joined
·
27,533 Posts
Quote:
Originally Posted by billbartuska View Post

If you use water temp (only) to control fan speeds the fans will speed up and slow down based on ambient temps, as that's what the water does. On cool days your fans will run slower and on warm days you fans will run faster at the same CPU/GPU loads. Not really what we want. Controlling Delta T controls how much heat is actually being removed.

If you made a chart that relates Delta Ts to CPU/GPU temps at a constant load and varying ambient temps (using IBT, as it more accurately represents real world full loads as compared to Prime95 or OCCT) and use that to manually set fan speeds, then fans would always run at the minimum speed required to supply the amount of cooling needed.
But that's way to complicated and basically, the only thing it really accomplishes is:
1.) To save a small amount of electricity (compared to just changing fans from 5v to 12v as needed)
2.) To run the fans quieter (sometimes).
Honestly, I see no problem with fans running faster on hot days. To me it's the same as fans running faster when system is working harder. It is what I expect the fans (and possibly puump) to do to maintain component temp .. as that is the important thing and why fan speed changes.

The only reason fans run faster on hot days and slower on cool days is because the radiators cool better with cooler air and therefore don't need the fans to run as fast. The water temps are the same, therefore the CPU and /or GPU will be the same (or very close) regardless of the air temperature. This of course assume the system has enough cooling capacity to maintain a constant cooling level.

Maybe we are not understanding what the other is saying here.
If I set fan speed to maintain a constant water temp, than CPU / GPU will maintain a constant temp as well.

I used the car analogy because it is easy to understand how the engine maintains a near constant temp for idle to full load in arctic or desert climates .. assuming it is working properly. Granted there is a water flow thermostat to enable faster warm-up and keep engine within operating range in arctic conditions, but we are not dealing with those extremes with our computers.
wink.gif
 

·
Registered
Joined
·
174 Posts
Discussion Starter · #9 ·
We've mild summers and winters and an old house with massive walls. Generally speaking indoor ambient doesn't fluctuate a lot here. So I'm fine with the correlation of ambient temperature and fan speed.

My current thought is to go for the Aquaero variant, connect the fans and inline water sensor(s) to it and use Aquasuite to control everything. This way I still have control of how much air I'm moving depending on what I'm doing (Which is generally gaming). At the same time I should be able to circumvent a possible issue with PWM signal strength from the MB header.
 

·
Overclock Failed...
Joined
·
13,565 Posts
Quote:
Originally Posted by doyll View Post

Maybe we are not understanding what the other is saying here.
If I set fan speed to maintain a constant water temp, than CPU / GPU will maintain a constant temp as well.
Oh, we're on the same page here.
But that's am interesting WCing system you've got there if the air going into the rads changes from 68°f to 95°f and the water temp doesn't change. Kind of sidesteps the Second Law of Thermodynamics.
 

·
Premium Member
Joined
·
27,533 Posts
Quote:
Originally Posted by billbartuska View Post

Oh, we're on the same page here.
But that's am interesting WCing system you've got there if the air going into the rads changes from 68°f to 95°f and the water temp doesn't change. Kind of sidesteps the Second Law of Thermodynamics.
I doubt it.
You are now talking stupid.
Get real.
If the air temp rises from 68f to 95f than the radiator will not be cooling the same .. meaning the water will be hotter .. means the fans will speed up to increase the cooling and lower the water temp.

Just like the cooling on a car with electric fans on radiator. Take a look. There is a sensor on the radiator monitoring the water temp and controlling the fan speed. Higher water temps mean higher fan speed.
There is also a temperature sensor on the engine itself.

While the engine does have sensors for barometric pressure. air temperature, oxygen level, etc., they are all used for the fuel mixture, ignition timing, etc. to control exhaust emissions and power efficiency.
 

·
Registered
Joined
·
174 Posts
Discussion Starter · #12 ·
Is the prime issue discussed then the lag/difference occuring between an increase in CPU, GPU, water and ambient temperature?

If so, would have having a sensor that picks up ambient temperature be an option for one rad and its fans and another rad with its fans for a water sensor? Could this be used to bridge the gap between CPU/GPU (and ultimately water) temperature increase and changes in (intake) ambient temperature?
 

·
Premium Member
Joined
·
27,533 Posts
Quote:
Originally Posted by Domiro View Post

Is the prime issue discussed then the lag/difference occuring between an increase in CPU, GPU, water and ambient temperature?

If so, would have having a sensor that picks up ambient temperature be an option for one rad and its fans and another rad with its fans for a water sensor? Could this be used to bridge the gap between CPU/GPU (and ultimately water) temperature increase and changes in (intake) ambient temperature?
You can, but why?
Room ambient raises and falls rather slowly. Much slower than water flow form blocks. So as long as fans change speed as needed when CPU / GPU temps change, they will stay cool .. up to whatever limit their waterbloacks, pumps, radiators and fans are capable of.
 

·
Registered
Joined
·
174 Posts
Discussion Starter · #14 ·
I'm trying to determine the why
wink.gif


Say I run one rad on a water sensor, is it beneficial to run the other rad on another sensor (CPU perhaps) or are the two spots after blocks (CPU and GPU) the biggest thermal indicators?

Reason for asking is because general consencus seems that the order of blocks and such doesn't matter (barring res to pump) much and I was planning to have the CPU and GPU blocks right after one another.
 

·
Overclock Failed...
Joined
·
13,565 Posts
Quote:
Originally Posted by doyll View Post

Maybe we are not understanding what the other is saying here.
If I set fan speed to maintain a constant water temp, than CPU / GPU will maintain a constant temp as well.
Quote:
Originally Posted by doyll View Post

[If the air temp rises from 68f to 95f than the radiator will not be cooling the same .. meaning the water will be hotter.
I took your first quoted remark to mean that if you didn't change the fan speeds the CPU/GPU temps would remain the same.
 

·
Premium Member
Joined
·
27,533 Posts
Quote:
Originally Posted by Domiro View Post

I'm trying to determine the why
wink.gif


Say I run one rad on a water sensor, is it beneficial to run the other rad on another sensor (CPU perhaps) or are the two spots after blocks (CPU and GPU) the biggest thermal indicators?

Reason for asking is because general consencus seems that the order of blocks and such doesn't matter (barring res to pump) much and I was planning to have the CPU and GPU blocks right after one another.
The hottest water will be coming out of the CPU or GPU blocks. Unless radiators are on different loops the water will get progressively cooler the farther it goes. Now if you have a block to radiator to block to radiator to pump loop, than I would monitor each block for the fans on radiator after the block. This will control the fans to get the best cooling before water goes to next block.
 

·
Premium Member
Joined
·
27,533 Posts
Quote:
Originally Posted by billbartuska View Post

I took your first quoted remark to mean that if you didn't change the fan speeds the CPU/GPU temps would remain the same.
Ah.. okay.
thumb.gif

fan speed to maintain a constant water temp.
Fan speed is not a fixed rpm. I should have said fan speed curve. Sorry about that.

But
in my defense
The discussion was about fan speeds changing .. and I was replying to your statement about fans running faster on hot days than cool days.
My opening paragraph said
Quote:
Honestly, I see no problem with fans running faster on hot days. To me it's the same as fans running faster when system is working harder. It is what I expect the fans (and possibly puump) to do to maintain component temp .. as that is the important thing and why fan speed changes.
so pretty obvious I was not talking about fixed speed fans.
biggrin.gif
 

·
Registered
Joined
·
174 Posts
Discussion Starter · #18 ·
Alright, that's settled then. Depending on loop 1 or 2 inline water sensors connected to an Aquaero along with the fans. No issue with PWM signal strength from the MB PWM header that way. Only fans left to pick.

Thanks to both of you, appreciate the discussion.
 
1 - 18 of 18 Posts
Top